Measuring walking speed on a "White Cane"

Hello everyone!

Im currently working on a research project where I am designing a white cane for visually impaired students. Currently the cane measures and records the swing speed as well as the angle that the cane in held by the person. Now I need to record the speed the user in walking at.

What I have considered already:

--I can't use GPS due to the location of use being mostly indoors and my professor doesn't want me to even try it.

-- I thought about using an IMU but the integration over and over will eventually just give me a large error I would think.

-- Also considered another accelerometer and just approximating the time between steps and averaging it with the users average step length.

--Built an encoder as well that went on the tip of the cane but with the swinging back and forth it and taps of various lengths it was not a feasible option.

I'm not sure what else to try at this point. I am thinking the accelerometer approximation method might be the best bet but I wanted to see if I could get some outside opinions before I fully commit to this.

Also considered another accelerometer and just approximating the time between steps and averaging it with the users average step length.

This is the only option (among your suggestions) that has a chance of working.

However, an ultrasonic ranger could ping in the forward direction, and if obstacles are encountered, both distance and velocity can be estimated. Range is up to 10 meters for the best of them.

Integrating IMU or accelerometer data is hopeless, as you can read about here.

jremington:
However, an ultrasonic ranger could ping in the forward direction, and if obstacles are encountered, both distance and velocity can be estimated. Range is up to 10 meters for the best of them.

Thinking this would get poor results as well if used in a crowded area with people moving all about. For example, someone could walk toward the sensor while the subject is stationary and skew the results. Just thinking.

Thank you for the replies. I considered the ultrasonic thing too and with the swinging of the cane plus the people/objects around the school I think my readings would be skewed.

I'm thinking the accelerometer method might be my only option that isn't over the top right now.

MPU 9050 or ICM 20648 have step count registers, although I don't know their accuracy and whether a decent documentation is really available for these products.

I think the pedometer is the only practical solution if you're limited to sensors on the cane+person. Most people with FitBit or other similar devices seem to like them and believe the distance they give.

What are trying to do with the output data, there may be other techniques to combine with the simple cane motion.

Optical mouse sensor (with replacement optics) in the stick's tip?

Sorry for the late response. My senior year and many projects are currently going on!

The output data is just to determine general speed. The instructors would like to know how the walking speed correlates with the swing speed in familiar vs unfamiliar settings.

I will have to look more into the optical sensor. I briefly considered something like that but I was not sure about the accuracy given the distance from the tip to the ground would be varying constantly.

For making a pedometer, do you guys think a simple single axis accelerometer would suffice? My thought was just one in the handle measuring the up/down z axis would work since that should be triggered when a step occurs.
The other option would be a second 3 axis accelerometer and measure up/down plus forward acceleration and only count steps when both forward motion and up/down motion occur.

This could go quite complex...
As you have observed, you have no idea what the surface of the terrain is at any given moment, plus steps, incline etc.

Your last thought seems to be heading along the right path (*no pun), but you can do all that with a single 3D accelerometer. The difficulty is still the accuracy of the 'step size'. It could be a doddery old fellow, that steps 12 inches per step, vs a younger user at 18-20 inches.

I'm sure there must be a way to bring all the data together for something... perhaps that nicluides the time taken between each pass of the walker's leg? Tied with angular velocity, etc.